Sharon Goldman

NEW: OpenAI gives first detailed debrief of the Hugging Face incident at Black Hat conference In a session I attended today at Black Hat, OpenAI's Eric Wallace and Michael Dalton said the company is "consciously slowing down research to enhance security" while a full technical postmortem is still underway. * OpenAI traced the roots of the attack back to May 7, during training of an unreleased frontier model—not July. * The most surprising detail: AI agents accidentally created an internal message board, allowing separate evaluation runs to share exploits, discoveries and work assignments. * OpenAI said it shut the message board down after an internal security incident—only for the agents to independently recreate it days later using a different communication method. * OpenAI called the incident a "watershed moment" for AI security and warned that "agent orchestrated fully automated offensive attacks are real now." * The company also said it is "consciously slowing down research to enhance security" while overhauling its defenses. groundlevel-ai.com/p/openai-…

Should I bother watching my a1c then? I will let my primary care know that my prediabetes likely won't have time to turn into diabetes anyway. Also seems like health insurers might want to know about this - premiums a bit of a waste?

But telling people to prepare for death by AI in three years is fine anyway?

😂😂😂

The Jensen news has me harking back to March 2024. At the packed SAP Center in San Jose, California, I was watching Huang give his keynote at Nvidia’s GTC developer conference — the first in-person GTC in five years and the first held in a venue that big. Wearing his signature black leather jacket, Huang joked that he hoped the audience realized they weren’t at a concert. But as the audience cheered and roared, it felt like one. In the year and a half since generative AI burst into the public consciousness with the release of ChatGPT, Nvidia, which supplied the vast majority of the chips powering LLMs, had become the biggest rock star of the AI boom, and Huang was its Mick Jagger. Three months later, Nvidia became the world’s most valuable company, when its market capitalization surpassed $3.3 trillion. I’d been following Nvidia’s rise closely since early 2023, when I wrote a deep dive on how the company came to dominate AI just as the generative AI boom took off. Executives, researchers and analysts walked me through Nvidia’s competitive advantages, from its CUDA software ecosystem to its willingness to make enormous bets on technologies years before they paid off. The conversations kept coming back to Huang himself, who had been called the “next Steve Jobs.” Two years after that night in San Jose, the only thing that had changed was Nvidia’s stock price, which kept climbing. This past March, Huang forecast a staggering $1 trillion in orders for Nvidia’s most advanced AI chips through 2027, driven by the AI infrastructure going up around the world. But at the time, I was struck by how Huang had also somehow avoided the public criticism aimed at other prominent AI CEOs, including OpenAI’s Sam Altman, Meta’s Mark Zuckerberg and Anthropic’s Dario Amodei. While social media railed against supposedly “evil” AI billionaires, Huang was still celebrated as the engineer-builder behind the boom. The huge AI data centers rising across the country have drawn fierce local opposition over electricity costs, water use and their effects on communities, and they are packed with Nvidia chips. Yet no one was protesting outside Nvidia’s Santa Clara headquarters. That changed this weekend — perhaps for good. groundlevel-ai.com/p/jensen-…

Is this the AI jump the shark moment 🦈

planning a trip to SF dec 2-5. Anything going on? ;)

TG-AI-F: The AI boom isn't over. But it may never be the same After a bruising week for AI stocks, the industry faces a new political reality. Plus, my experience with OpenAI’s Dots and what the company says about the economics of always-on AI agents. groundlevel-ai.com/p/tg-ai-f…

The AI data center boom is entering its bargaining era I checked back with data center opponents in Texas, Michigan and Louisiana after the midterms, then asked infrastructure experts what's next. The answers suggest a complicated road ahead.

I wrote one story for Salon back in the day. Fare thee well.

The AI data center boom is entering its bargaining era. Over the past year, I traveled to Texas, Michigan, Louisiana and Arizona to report on communities fighting massive AI data center projects. After Tuesday's election, I checked back in with some of the people I met to ask what they thought of the results. The reactions were mixed. In Texas, a woman whose family has farmed the same land for generations was happy to see Greg Abbott lose. In Michigan, a mother of five fighting the $16 billion OpenAI/Oracle Stargate project in court was skeptical about the new Democratic governor. And in Louisiana, a resident opposing Meta's Hyperion project was struck by how close the Senate race had been. Meanwhile, a developer with projects in Texas told me she isn't particularly worried. Projects that already have zoning and power lined up should be fine, she said. The more speculative ones could face delays. And OpenAI is already on a charm offensive, with a new framework promising to pay its own electricity costs, use less water and give communities more say. Meanwhile, in September 2025, I stood at OpenAI's Stargate site in Abilene as Ted Cruz declared, “Texas is ground zero for AI.” My latest for Ground Level AI: groundlevel-ai.com/p/the-ai-…